Original research paper
Prognostic value of electrically evoked auditory brainstem responses in cochlear implantation Karin Lundin, Fredrik Stillesjö, Helge Rask-Andersen Section of Otolaryngology, Department of Surgical Sciences, Uppsala University Hospital, Uppsala University, Sweden Objectives: The aim of this study was to investigate whether electrical auditory brainstem responses (eABRs) obtained during cochlear implantation (CI) can predict CI outcomes. We also aimed to assess whether eABR can be used to select patients for auditory brainstem implantation (ABI). Methods: This was a retrospective study. The latencies and quality of the eABR waveforms from adult patients implanted with CI in Uppsala from 2011 to 2013 (n = 74) and four children with severe cochlear abnormalities were analyzed. Speech perception was assessed through postoperative monosyllabic word (MS-word) recognition. A score was constructed for each patient based on wave II, III, and V patency. Results: eABR latencies increased towards base stimulation of the cochlea. Wave V for the mid- and lowfrequency regions was the most robust. Significant latency shifts occurred in wave V from the low- to highfrequency regions (**P < 0.01) and from the mid- to high-frequency regions (**P < 0.01). No correlations were found between waveform score, wave V–III interval, wave V latency, and MS-word scores. A negative eABR always predicted a negative outcome. Among the patients with negative outcomes, 75% had eABRs. Discussion: Implant electrical stimulation and brain stem recordings can be used (eABRs wave V) to predict a negative functional outcome. Low-frequency waves V were observed in all patients with successful CI outcomes. Patients for whom eABR waveforms were completely absent had unsuccessful CI outcomes. Keywords: Cochlear implant, Auditory brainstem implant, Electrical auditory brainstem response
Introduction Cochlear implantation (CI) is an effective treatment for adults and children with severe to profound hearing loss. Most patients benefit from the CI and improve in open-set speech perception. However, patient outcomes vary, probably owing to several factors, such as anatomical conditions, cognition, age at implantation, deafness duration, cause of deafness, and age at onset of deafness. The function and integrity of the central auditory pathway is also of great importance. In some children with severe inner ear malformations, uncertainties as to the presence of a patent auditory nerve may arise even following high-resolution magnetic resonance imaging (MRI) investigations. The question of whether to implant a patient with a CI or an auditory brain stem implant (ABI) can be difficult to answer. In these cases, an objective technique to establish a functionally Correspondence to: Karin Lundin, Section of Otolaryngology, Department of Surgical Sciences, Uppsala University Hospital, Uppsala University, SE-751 85 Uppsala, Sweden. E-mail: [email protected]
254
© W. S. Maney & Son Ltd 2015 DOI 10.1179/1754762815Y.0000000005
responsive neural pathway from the cochlea to the cochlear nucleus is important. Electrically evoked auditory brain stem responses (eABRs) can be used intra-operatively (Tysome et al., 2013). Cinar et al. (2011) compared patients with or without inner ear malformations and were able to record eABRs from 100% of the patients without malformations and 99.44% of those with inner ear malformations. Both authors (Cinar et al., 2011; Tysome et al., 2013) stated that the fact that eABRs could be recorded in patients with malformed ears was critical and demonstrates the reliability and superiority of eABR as an objective assessment technique. However, eABR latencies were found to be significantly longer in malformed inner ears when compared with ears without malformations. Battelino et al. (2009) found that in some cases a poor eABR response could predict a poor CI-outcome. Firszt et al. (2002b) found no relationship between speech perception with CI and eABR amplitude or latencies. On the other hand, Gallégo et al. (1998) found that wave V–III interval
Cochlear Implants International
2015
VOL.
16
NO.
5
Lundin et al.
was a strong predictor of CI outcomes, and Gibson et al. (2009) and Walton et al. (2008) stated that the quality of the eABR waveform appears to correlate well with postoperative speech perception. They assessed eABR waveform and found a significant correlation between its quality and speech perception with CI. Since 2011, eABR measurements have been routinely performed during CI surgery in Uppsala. Here, we analyzed the eABR waveform quality, latencies, and inter-peak intervals separately for electrodes covering the basal, middle, and apical parts of the cochlea from implanted adults. These responses were explored to assess their potential to predict CI outcomes in terms of speech perception as measured by monosyllabic (MS) word recognition. In addition, four children with expected cochlear nerve absence were included. Our aim was to investigate whether eABR can be used to predict CI outcomes and aid in ABI decision-making in pertinent cases.
Methods Patients All adult patients implanted in Uppsala from 2011 to 2013 in whom eABR measurements were intraoperatively measured were included in this study (n = 74). Mean age at implantation was 65 years (median 69 years, range 18–87). Four children were tested in this study including two with CHARGE syndrome (2 and 3 years old), one with VACTERL association (2 years old) and one with Goldenhar’s syndrome (2 years old). All children were investigated with highresolution MRI prior to implantation, and eABRs were documented as an additional measure to rule out a partially working hearing nerve.
Device description Implants from Cochlear (Lane Cove, Australia; CI24RE(CA), CI422, or Hybrid L24) were used in 37 implantations, and implants from MedEl (Innsbruck, Austria; Sonata 31 or 24, or Concerto 20, 24, 28, or 31) were used in 37 implantations in the adult group. Of the four children, two were implanted with an implant from Cochlear (CI24RE(CA) and CI422) and two were measured using a test electrode from MedEl, which is an 18 mm long simplified cochlear implant with three contacts in the cochlea.
eABR measurements All recordings were made during surgery directly after implant insertion. Two separate systems were used, one for recording and one for stimulation. The stimulation system used was the Cochlear or MedEl programming system, which uses the CI as a stimulator. The recording system was an evoked potentials (EP)
Prognostic value of electrically evoked auditory brainstem responses
system, which was triggered to record from the stimulation system. Two different recording systems were used at the clinic, the Otometrics Chart 200 (GN Otometrics, Taastrup, Denmark) and the GSI Audera (GSI, Minneapolis, USA). Recording needle electrodes were positioned at the Vertex (Act), C7 (Ref ), and at the hairline on the neck (ground). Three different electrodes were stimulated for each patient, one each in the low-, mid-, and high-frequency regions. The stimulation pulse width was 25 μs, and the current level was 235 CL for Cochlear devices and 30 μs and 1000 q.u. for MedEl devices. These conditions produced the same stimulus charge from both systems and were above most patients’ C-levels in their sound processor maps. The recording system averaged 1000 sweeps and was filtered by a low pass filter at 3 kHz and a high pass filter at 10 Hz (GSI Audera) or by a low pass filter at 5 kHz and a high pass filter at 5 Hz (Otometrics Chartr 200). Latencies were measured on the result from an alternating stimulation or as an average of the latencies from the positive and negative stimulations in cases where no alternating polarity was used. In most cases, the eABR responses were verified by inverting the polarity of the stimulation, which resulted in an inversion of the artifact without an inversion of the eABR waveform. Figure 1 shows a typical result from the different modes of stimulation. The most identifiable peaks (II, III, and V) were marked according to standard clinical procedure.
Waveform assessment A modified classification table as described by Gibson et al. (2009) and Walton et al. (2008) was used to classify the quality of eABR waveforms (Table 1). We chose to simplify the table by including only latencies and leaving out amplitude measures, because we stimulated only at a single level, and both latencies and amplitudes change according to the level of stimulation. Abbas and Brown (1988) and Kim et al. (2008) claim that amplitudes undergo most changes when the stimulation level is modified, whereas latency changes are relatively small. A score of 3 represents a superior eABR quality and 0 an absent eABR response. A total waveform score was constructed by adding the scores from the three measured electrodes (0–9).
Hearing evaluation MS-words (Svensk Talaudiometri, C-A Tegnér AB, Stockholm, Sweden, 1998) were used to test speech perception in a sound-treated booth in free field at 65 dB SPL with the loud speaker at 0° azimuth. Only postoperative MS-word scores with CI were analyzed. In patients with no speech recognition (0% in MS-word test) bi-syllabic word testing and the threedigit test (Svensk Talaudiometri, C-A Tegnér AB,
Cochlear Implants International
2015
VOL.
16
NO.
5
255
Lundin et al.
Prognostic value of electrically evoked auditory brainstem responses
Table 1 Implant evoked electrical auditory brain stem response waveform score; table modified from Gibson et al. (2009) and Walton et al. (2008) Presence of waves Score
eII (Y/N)
eIII (Y/N)
eV (Y/N)
3 2 1 0
Y N N N
Y Y N N
Y Y Y N
compare latencies between wave V at the three different locations (base, middle, apex), and Bonferroni adjustments for three repeated measures were made. The Wilcoxon signed-rank test was used to compare wave form scores for the three different locations (base, middle, apex), and Bonferroni adjustments for three repeated measures were made. Linear regression analysis was used to explore relationships between waveform scores, waveform V latencies, wave V–III interval and MS-words. For waveform V latencies and wave V–III intervals, Pearson’s correlation coefficient (r) was calculated and for waveform scores, Spearman’s rank correlation coefficient (rs) was calculated.
Results eABR latencies and assessment of waveform quality
Figure 1 Typical eABR waveforms. (A) Recordings from alternating, positive and negative stimulation modes and (B) recordings from positive and negative modes.
Stockholm, Sweden, 1998) were also performed. Speech perception data were collected at 6 months or 1 year after surgery.
Statistical analyses Statistical analyses were made using SPSS 22.0 software (IBM 1989, 2013). Paired t-tests were used to
Latencies for wave II, III, and V are displayed in Table 2. Latencies increased towards the base of the cochlea. Significant latency differences were found between wave V in the low-frequency region and the high-frequency region (**P < 0.01, 95% confidence interval of the mean difference 0.09–0.46 ms), and between wave V in the mid-frequency region and the high-frequency region (**P < 0.01, 95% confidence interval of the mean difference 0.06–0.33 ms). Wave V values for the mid-frequency and the low-frequency regions were the most robust waveforms. In 66 of 74 (89%) recordings in the low-frequency region, we were able to identify wave V. In eight patients, wave V could not be observed. In five of these cases, the measurement failed owing to technical reasons, in two cases, the measurement was not performed on
Table 2 Means, standard deviations (SD) and 95% confidence intervals for eABR waves II, III and V. Latencies in ms. Ninety-five percent confidence interval for means in parentheses Low-frequency region Mean
SD
Wave II latency (ms) 1.42 (1.36–1.48) 0.17 Wave III latency (ms) 2.17 (2.11–2.22) 0.21 Wave V latency (ms) 3.99 (3.90–4.07) 0.36
256
Mid-frequency region
High-frequency region
n
Median
Mean
SD
n
Median
Mean
SD
n
Median
32
1.42
1.45 (1.35–1.54)
0.25
28
1.42
1.46 (1.31–1.60)
0.23
12
1.43
57
2.12
2.23 (2.17–2.29)
0.22
56
2.20
2.28 (2.17–2.39)
0.25
22
2.29
66
3.98
4.15 (4.06–4.24)
0.35
64
4.14
4.32 (4.17–4.48)
0.41
29
4.42
Cochlear Implants International
2015
VOL.
16
NO.
5
Lundin et al.
Prognostic value of electrically evoked auditory brainstem responses
Table 3 Mean scores and mean wave V–III differences; 95% confidence intervals for mean in parentheses.
Mean score (0–3) Median score (0–3) SD n Mean wave V–III (ms) Median wave V–III (ms) SD n
Low-frequency region
Mid-frequency region
High-frequency region
2.3 (2.14–2.51) 2 0.8 67 1.78 (1.70–1.86) 1.79 0.28 57
2.1 (1.92–2.34) 2 0.9 69 1.89 (1.82–1.96) 1.86 0.25 56
1.0 (0.66–1.25) 0 1.2 62 1.94 (1.82–2.06) 2.00 0.26 20
those electrodes, and in one case, wave V was not detectable. Therefore, in all but 1 of the 67 successful recordings (99%), wave V was detectable for the lowfrequency region. In the mid-frequency region, we identified wave V in 64 of 74 (86%) recordings. In 10 cases, wave V was not observed. In five of these cases, the measurement failed owing to technical reasons, and in the other five cases, wave V was not detectable. Therefore, in all but 5 of the 69 successful recordings (93%), wave V was detectable for the middle region. The waveform quality as assessed by a score from 0 to 3, was highest at the low- and midfrequency regions of the cochlea (mean scores 2.3 and 2.1, respectively) compared with 1.0 in the highfrequency region. There was a statistically significant difference between the waveform scores for the low and mid-frequency regions of the cochlea compared with the high-frequency region (***P < 0.001). Wave V–III latency increased towards the high-frequency region (Table 3). It was not possible to fully evaluate how the difference in electrode lengths contributed to
the waveform score, because the number of patients in each implant length group varied essentially. The mean waveform score was highest for the longest electrode (31 mm) though.
Hearing evaluation There were no correlations between MS-word scores and wave V latency in the three different regions (low r = −0.1 p = 0.4, mid r = −0.2 P = 0.07, high r = −0.1 P = 0.6) or wave V–III intervals in the three different regions (low r = −0.05 P = 1.0, mid r = −0.2 P = 0.2, high r = 0.04 P = 0.9) using linear regression analysis (Fig. 2). We also did not find any correlation between MS-word scores and the total score (rs = 0.2, P = 0.3) or the scores in the three different regions (low rs = −0.07, P = 1.0, mid rs = 0.06, P = 0.7, high rs = 0.2, P = 0.3) using linear regression analysis. No adjustments for repeated measures were performed, because no correlations were found.
Figure 2 Correlation between wave V latencies and monosyllabic (MS) words with CI. (A) Low frequencies (r = −0.1, P = 0.4) (B) mid frequencies (r = −0.2, P = 0.07) (C) high frequencies (r = −0.1, P = 0.6) and correlation between wave V–III interval and MSwords with CI (D) Low frequencies (r = −0.05, P = 1.0) (E) mid frequencies (r = −0.2, P = 0.2) (F) high frequencies (r = 0.1, P = 0.6).
Cochlear Implants International
2015
VOL.
16
NO.
5
257
Lundin et al.
Prognostic value of electrically evoked auditory brainstem responses
Table 4 Patients with absent wave V in the middle and/or apex regions and patients with no speech discrimination (MS-words, bi-syllabic words and three-digit-test)
Patient no.
Wave V (lowfrequency region)
Wave V (midfrequency region)
Total wave form score (0–9)
Patients with absence of wave V in the middle and/or apex region 12 No No 0
22
Yes
No
2
31
Yes
No
1
41
Yes
No
2
55
Not measured
No
–
Patients with no speech discrimination 9 Yes Yes
9
12
No
No
0
48
Yes
Yes
5
49
Yes
Yes
6
55
Not measured
No
–
74
Yes
Yes
4
Cochlear Implants International
2015
VOL.
16
NO.
Patient has very high levels in the map postoperatively. Detects only sound, no word discrimination. Acoustic neuroma in the CI-ear. No speech discrimination. Implant: Concerto28 (28 mm) Patient with otosclerosis. Sudden loss of hearing. Contralateral ear deaf for 30 years. Patient in coma for five weeks owing to Legionella one year before CI-surgery. Twentyfive percent on auditory only three-digit-test. Implant: CI422 (25 mm) Patient with otosclerosis and high levels in the map postoperative. Twenty-four percent on MS-word test. Implant: CI422 (25 mm) Young patient (18 years old at surgery) with high-frequency deafness. Fifty-eight percent on MS-word test. Implant: Concerto24 (24 mm) Patient reports mainly a feeling from implant stimulation postoperatively. Long-time deafness (50 years) before surgery. No speech discrimination. Implant: Concerto31 (31 mm) Deaf for 72 years prior to CI. CI use gives severe tinnitus. Non-user. Implant: Concerto31 (31 mm) Patient has very high levels in the map postoperatively. Detects only sound, no word discrimination. Acoustic neuroma in the CI-ear. Uses CI full time for environmental sounds. Implant: Concerto28 (28 mm) Reports sound and a feeling from implant use. Deaf for 37 years in the CI-ear prior to CI. Non-user. Implant: CI422 (25 mm) Severe prelingual hearing loss. 50 years old at implantation. Uses CI full time for environmental sounds. Implant: CI422 (25 mm) Patient reports mainly a feeling from implant stimulation postoperatively. Deaf for 50 years prior to CI. Uses CI but considers stopping using it. Implant: Concerto31 (31 mm) Not measured. Do not show up to appointments. User? Implant: Hybrid L (17 mm)
Table 4 shows the cases in which wave V was not detected for reasons others than technical failures at the mid- and/or low-frequency regions, and the cases in which patients had no open-set speech discrimination on any of the performed speech tests. In addition, the electrode lengths are shown in Table 4. A negative outcome was defined as a patient with no recordable speech discrimination with CI (measured as scoring 0% on MS-words or less than 10% on bi-syllabic words or less than 25% on the three-digit test). The results showed that if the patient has no detectable wave V in the low-frequency region, the outcome will most certainly be unsuccessful. Patients with no speech perception with CI had eABR wave V in the low-frequency region in 75% of the cases. One remarkable case with a total waveform score 9 and deafness duration before CI surgery of 72 years received hearing sensations from the implant (no speech perception) but could not use it, because the implant induced severe tinnitus.
258
Comment
5
Children with cochlear abnormalities Among the four children, no eABR waveforms could be seen from any of the stimulations (total score 0). Here, stimulation levels were increased to determine that no waveforms occurred at higher levels. One child tested via a Nucleus CI24RE(CA) implant received an ABI during the same surgery. In one child, the surgery proceeded with the intention to implant an ABI but had to be terminated for surgical reasons. One child did not receive an implant as parents and medical staff decided before surgery in case eABR measurements were negative. One child was implanted and monitored by a Nucleus CI422 implant and programmed 4 weeks after surgery. The child used the implant for 6 months but did not show any reactions to sound even though levels in the CI were set at the maximum.
Discussion Although interference and electrical noise from other surgical equipment may sometimes occur, eABR
Lundin et al.
recordings via the CI are relatively straightforward during surgery in most patients. Recording electrode placement and marking/coupling of the lead-wires to the EP equipment may be crucial factors. We found that wave V for mid-frequency and lowfrequency regions were often robust and wave V for the low-frequency region were recorded in every case where outcomes was successful, i.e., in patients benefiting from the implant. However, those waveforms were also obtained in a few cases when outcome was not successful (Table 4). Walton et al. (2008) described two children with auditory neuropathy spectrum disorder (ANSD) that upon MRI showed absence of auditory nerve but still presented recordable eABR responses via the CI. Both children had a poor CI outcome. In the three cases, in our study where no eABR-waveform responses were obtained ( patient no. 12 and 55, Table 4) and the child that received an implant despite absent intra-operative eABRs responses, the CI outcomes were unsuccessful and patients did not demonstrate open-set speech discrimination. However, the two adult patients continued to use their implants and benefitted from receiving background sound and/or awareness from the implant. Recently, one additional patient who showed no eABR responses was also implanted with an unsuccessful outcome. One crucial question is if there is possible to benefit from CI despite no eABR responses. Jeon et al. (2013) found that patients with ANSD did not always display eABR responses but had good CI outcomes. Greisiger et al. (2011) investigated eight children with ANSD and in comparison with eight non-ANSD children found no differences in eABR amplitudes or waveforms between the two groups. These authors found patent wave III and V in all investigated patients. Walton et al. (2008) compared eABR waveform quality for children with and without ANSD. The children with ANSD had significantly worse CI outcomes as well as decreased waveform quality compared with the nonANSD children. They also found a significant correlation between waveform quality and CI outcomes. Our results show longer latencies and indistinct waveforms at the high-frequency region compared with those of the low-frequency regions of the cochlea are consistent with earlier studies (Firszt et al., 2002a; Gordon et al., 2007; Miller et al., 1993). Zimmerman et al. (1995) and Gordon et al. (2007) showed that humans with hearing loss display smaller populations of spiral ganglion cells in the cochlear base (high-frequency region) compared with the apex (low-frequency region). Because the number of patients using a specific electrode length varied extensively in this study, no statistical analysis was performed regarding the influence of different electrode lengths on waveform scores. The length, size, and
Prognostic value of electrically evoked auditory brainstem responses
shape of the cochlea also vary greatly between individuals. Erixon et al. (2009) concluded in their study of 73 adult cochleae that the length of the first turn could vary from 20.3 to 24.3 mm. We could observe though that the mean total waveform score was higher for the longest electrodes (31 mm). Furthermore, Gordon et al. (2007) showed that latencies shortened after a year of implant use for all waves (II–V) at both high- and low-frequency regions in 50 examined children with early onset of severe to profound deafness. Apical versus basal differences persisted with one exception; wave V–III interval differences decreased with longer implant use. Firszt et al. (2002a) found wave V–III intervals to be 1.62 ms across all electrodes (low, mid, high). They stated that eABR wave V–III intervals are shorter than acoustic ABR V–III intervals because electrical stimulation via the CI actually enhances neural synchrony. Our results showed longer wave V–III intervals compared with these authors (Firszt et al. 2002a): 1.89 ms for mid-frequency regions and 1.87 ms for all electrodes (low, mid, high). These values may be considered normal for acoustic ABR V–III intervals. One explanation for the longer interval in our study could be differences in interpretation of wave V. Wave V often consists of wave IV and wave V combined, and there is rarely a clear separation between these two waves. This lack of clear visualization makes wave V difficult to mark. Similar to Firszt et al. (2002b), we found no correlation between speech perception outcomes (MSwords) and waveform latencies or waveform quality. However, absent waveform predicted no or poor CI outcomes. Walton et al. (2008) and Gibson et al. (2009) found a positive correlation between waveform quality and the Melbourne speech perception score (MSPS). The MSPS consists of a seven-point scale and a minimum of score 4 is needed to develop speech perception. They assessed eABR responses on 22 electrodes, and the maximum total score obtained was 66. Walton et al. (2008) found that patients with eABR scores >56 are more likely to get a MSPS ≥ 4. This study assessed eABR only on three electrodes and the MS-word test was used to assess speech perception outcomes. The reason for measuring eABR on only three electrodes in this study was that measurements were an integral part of the regular clinical intra-operative CI test protocol, with the intention not to prolong surgery. One explanation for the different results could be the different speech perception measures used. Gallégo et al. (1998) found the wave V–III interval to be a strong predictor by testing 17 patients using a phoneme-test. They measured eABR at eight different levels to correct for the different postoperative levels required by patients. They also used an algorithm for automatic wave-latency measurement.
Cochlear Implants International
2015
VOL.
16
NO.
5
259
Lundin et al.
Prognostic value of electrically evoked auditory brainstem responses
Our study used a different speech perception score and would probably have been more accurate if we had been able to record from all electrodes at a number of different levels that could be compared at a later time as waveform latency decreases and amplitude increases as the stimulation levels are augmented. Battelino et al. (2009) performed eABR prior to CI using a golf club stimulating electrode close to the round window. They were able to record eABR in only 64% of their cases using that technique and measured CI outcome with pure tones audiometry after CI. In their 46 patients, a positive eABR strongly correlated with a positive CI outcome. We obtained similar results, because all patients who showed a positive eABR were able to detect sound through the implant although speech perception varied among these patients. Kubo et al. (2001) showed that the eABR amplitude growth curve correlated well with consonant recognition scores in cochlear implant patients 1 month following surgery, but when tested at 3, 6, and 12 months after implantation, the correlation was no longer apparent. They also concluded that the event-related potential (P300) latency was a better indicator for hearing ability after CI, meaning that the higher auditory system is of greater importance for speech perception with CI. To use a CI for eABR in the decision-making process between CI and ABI, as done for two of the four children in this study, may not be considered cost-beneficial. The reason for this was the uncertainty of the radiologist to establish the presence of a patent or hypoplastic auditory nerve. Therefore, it was considered necessary to first try a CI. An alternative was to perform round window electrode stimulation. Owing to particular conditions it was found better to proceed with a CI testing before continuing the ABI surgery. In the other two pediatric cases, eABRs were assessed using a test electrode from MedEl (Innsbruck, Austria). It consists of an 18 mm long basic CI array with three contacts inside the cochlea. Such a device may be more useful and cost-effective in future decision-making between CI and ABI. However, we believe that an even longer device than the current version might be beneficial. Not all measurements were performed in the alternating polarity in this study. This is a limitation as there is a slight shift in latencies between the different stimulation modes, even though the mean values were used in those cases. In addition, waveforms II, III and V were marked manually. A more accurate latency measure would have been obtained if we would have been able to stimulate at a number of different levels to adjust for the patients’ different levels used later on in their sound processor. Sometimes, eABR measurements were difficult to collect at the operating room for reasons such as interference during wound closure, etc.
260
Cochlear Implants International
2015
VOL.
16
NO.
5
Taken together, eABR monitoring seems to be a useful tool to further predict the outcome of CI surgery. In difficult cases with inner ear malformations as well as long-time deafness, eABR may be helpful to the surgeon and team to assess whether a patent or hypoplastic auditory nerve may provide sufficient electric information to the auditory cortex. However, several factors such as anatomical conditions, cognition, age at implantation, deafness duration, cause of deafness, and age at onset of deafness contribute to the final outcome and eABR responses cannot be used as sole predictors. However, we believe that this method gives important additional information and seems to have a definitive role in the diagnostic armamentarium to better consider complementary surgical strategies such as ABI.
Conclusion eABR via the CI is relatively straightforward during surgery in most patients. Wave V for the low-frequency regions was registered in all patients with successful CI outcomes. Patients with no eABR waveforms showed poor CI outcomes in terms of speech perception. eABR may be particularly useful in many cases to assist surgeons in strategically choosing between CI and ABI.
Acknowledgements The authors thank statistician Lars Lindhagen for advice on the data analyses.
Disclaimer statements Contributors KL is the corresponding author and has performed the measurements, analyzed the data, and written the text. FS has also performed the measurements and contributed to the writing and the analyzes. HR-A has done a massive contribution to the text. Funding This study was supported by grants from Gunnar Arnbrinks foundation and Uppsala läns landstings FoU-medel. Conflicts of interest The authors report no conflicts of interest. Ethics approval The study was approved by Uppsala Ethical Review Board (19/11-2014, Dnr: 2014/437).
References Abbas, P.J., Brown, C.J. 1988. Electrically evoked brainstem potentials in cochlear implant patients with multi-electrode stimulation. Hearing Research, 36: 153–162. Battelino, S., Gros, A., Butinar, D.A. 2009. Comparison of electroaudiometry (EAM) and electrically evoked brainstem response (eABR) with the cochlear implantation hearing results. International Advanced Otology, 5(1): 100–103. Cinar, B.C., Atas, A., Sennaroglu, G., Sennaroglu, L. 2011. Evaluation of objective test techniques in cochlear implant users with inner ear malformations. Otology Neurotology, 32: 1065–1074.
Lundin et al.
Erixon, E., Högstorp, H., Wadin, K., Rask-Andersen, H. 2009. Variational anatomy of the human cochlea: implications for cochlear implantation. Otology Neurotology, 30: 14–22. Firszt, J.B., Chambers, R.D., Kraus, N., Reeder, R.M. 2002a. Neurophysiology of cochlear implant users I: effects of stimulus current level and electrode site on the electrical ABR, MLR and N1-P2 response. Ear Hearing, 23(6): 502–515. Firszt, J.B., Chambers, R.D., Kraus, N. 2002b. Neurophysiology of cochlear implant users II: comparison among speech perception, dynamic range, and physiological measures. Ear Hearing, 23(6): 516–531. Gallégo, S., Frachet, B., Micheyl, C., Truy, E., Collet, L. 1998. Cochlear implant performance and electrically-evoked auditory brain-stem response characteristics. Electroencephalography Clinical Neurophysiology, 108: 521–525. Gibson, W.P.R., Sanli, H., Psarros, C. 2009. The use of intra-operative electrical auditory brainstem responses to predict the speech perception outcome after cochlear implantation. Cochlear Implants International, 10(S1): 53–57. Gordon, K.A., Papsin, B.C., Harrison, R.V. 2007. Auditory brainstem activity and development evoked by apical versus basal cochlear implant electrode stimulation in children. Clinical Neurophysiology, 118: 1671–1684. Greisiger, R., Tvete, O., Shallop, J., Elle, O.J., Hol, P.K., Jablonski, G.E. 2011. Cochlear implant-evoked electrical auditory brainstem responses during surgery in patients with auditory neuropathy spectrum disorders. Cochlear Implants International, 12(S1): 58–60.
Prognostic value of electrically evoked auditory brainstem responses
Jeon, J.H., Bae, M.R., Song, M.H., Noh, S.H., Choi, K.H., Choi, J.Y. 2013. Relationship between electrically evoked auditory brainstem response and auditory performance after cochlear implant in patients with auditory neuropathy spectrum disorder. Otology Neurotology, 34: 1261–1266. Kim, A.H., Kileny, R.R., Arts, P.R., El-Kashlan, H.K., Telian, S.A., Zwolan, T.A. 2008. Role of electrically evoked auditory brainstem response in cochlear implantation of children with inner ear malformations. Otology Neurotology, 29: 626–634. Kubo, T., Yamamoto, T., Iwaki, M., Matsukawa, M., Doi, K., Tamura, M. 2001. Significance of auditory evoked responses (eABR and P300) in cochlear implant subjects. Acta Otolaryngology, 121: 257–261. Miller, C.A., Abbas, P.J., Brown, C.J. 1993. Electrically evoked auditory brainstem response to stimulation of different sites in the cochlea. Hearing Research, 66: 130–142. Tysome, J.R., Axon, P.R., Donnelly, N.P., Evans, G. D., Ferner R.E., Fitzgerald O’Connor A.F. et al. 2013. English consensus protocol evaluating candidacy for auditory brainstem and cochlear implantation in neurofibromatosis type 2. Otology Neurotology, 34: 1743–1747. Walton, J., Gibson, W.P.R., Sanli, H., Prelog, K. 2008. Predicting cochlear implant outcomes in children with auditory neuropathy. Otology Neurotology, 29: 302–309. Zimmerman, C.E., Burgess, B.J., Nadol, J.B. 1995. Patterns of degeneration in the human cochlear nerve. Hearing Research, 90: 192–201.
Cochlear Implants International
2015
VOL.
16
NO.
5
261
Prognostic value of electrically evoked auditory brainstem responses in cochlear implantation Karin Lundin, Fredrik Stillesjö, Helge Rask-Andersen Section of Otolaryngology, Department of Surgical Sciences, Uppsala University Hospital, Uppsala University, Sweden Objectives: The aim of this study was to investigate whether electrical auditory brainstem responses (eABRs) obtained during cochlear implantation (CI) can predict CI outcomes. We also aimed to assess whether eABR can be used to select patients for auditory brainstem implantation (ABI). Methods: This was a retrospective study. The latencies and quality of the eABR waveforms from adult patients implanted with CI in Uppsala from 2011 to 2013 (n = 74) and four children with severe cochlear abnormalities were analyzed. Speech perception was assessed through postoperative monosyllabic word (MS-word) recognition. A score was constructed for each patient based on wave II, III, and V patency. Results: eABR latencies increased towards base stimulation of the cochlea. Wave V for the mid- and lowfrequency regions was the most robust. Significant latency shifts occurred in wave V from the low- to highfrequency regions (**P < 0.01) and from the mid- to high-frequency regions (**P < 0.01). No correlations were found between waveform score, wave V–III interval, wave V latency, and MS-word scores. A negative eABR always predicted a negative outcome. Among the patients with negative outcomes, 75% had eABRs. Discussion: Implant electrical stimulation and brain stem recordings can be used (eABRs wave V) to predict a negative functional outcome. Low-frequency waves V were observed in all patients with successful CI outcomes. Patients for whom eABR waveforms were completely absent had unsuccessful CI outcomes. Keywords: Cochlear implant, Auditory brainstem implant, Electrical auditory brainstem response
Introduction Cochlear implantation (CI) is an effective treatment for adults and children with severe to profound hearing loss. Most patients benefit from the CI and improve in open-set speech perception. However, patient outcomes vary, probably owing to several factors, such as anatomical conditions, cognition, age at implantation, deafness duration, cause of deafness, and age at onset of deafness. The function and integrity of the central auditory pathway is also of great importance. In some children with severe inner ear malformations, uncertainties as to the presence of a patent auditory nerve may arise even following high-resolution magnetic resonance imaging (MRI) investigations. The question of whether to implant a patient with a CI or an auditory brain stem implant (ABI) can be difficult to answer. In these cases, an objective technique to establish a functionally Correspondence to: Karin Lundin, Section of Otolaryngology, Department of Surgical Sciences, Uppsala University Hospital, Uppsala University, SE-751 85 Uppsala, Sweden. E-mail: [email protected]
254
© W. S. Maney & Son Ltd 2015 DOI 10.1179/1754762815Y.0000000005
responsive neural pathway from the cochlea to the cochlear nucleus is important. Electrically evoked auditory brain stem responses (eABRs) can be used intra-operatively (Tysome et al., 2013). Cinar et al. (2011) compared patients with or without inner ear malformations and were able to record eABRs from 100% of the patients without malformations and 99.44% of those with inner ear malformations. Both authors (Cinar et al., 2011; Tysome et al., 2013) stated that the fact that eABRs could be recorded in patients with malformed ears was critical and demonstrates the reliability and superiority of eABR as an objective assessment technique. However, eABR latencies were found to be significantly longer in malformed inner ears when compared with ears without malformations. Battelino et al. (2009) found that in some cases a poor eABR response could predict a poor CI-outcome. Firszt et al. (2002b) found no relationship between speech perception with CI and eABR amplitude or latencies. On the other hand, Gallégo et al. (1998) found that wave V–III interval
Cochlear Implants International
2015
VOL.
16
NO.
5
Lundin et al.
was a strong predictor of CI outcomes, and Gibson et al. (2009) and Walton et al. (2008) stated that the quality of the eABR waveform appears to correlate well with postoperative speech perception. They assessed eABR waveform and found a significant correlation between its quality and speech perception with CI. Since 2011, eABR measurements have been routinely performed during CI surgery in Uppsala. Here, we analyzed the eABR waveform quality, latencies, and inter-peak intervals separately for electrodes covering the basal, middle, and apical parts of the cochlea from implanted adults. These responses were explored to assess their potential to predict CI outcomes in terms of speech perception as measured by monosyllabic (MS) word recognition. In addition, four children with expected cochlear nerve absence were included. Our aim was to investigate whether eABR can be used to predict CI outcomes and aid in ABI decision-making in pertinent cases.
Methods Patients All adult patients implanted in Uppsala from 2011 to 2013 in whom eABR measurements were intraoperatively measured were included in this study (n = 74). Mean age at implantation was 65 years (median 69 years, range 18–87). Four children were tested in this study including two with CHARGE syndrome (2 and 3 years old), one with VACTERL association (2 years old) and one with Goldenhar’s syndrome (2 years old). All children were investigated with highresolution MRI prior to implantation, and eABRs were documented as an additional measure to rule out a partially working hearing nerve.
Device description Implants from Cochlear (Lane Cove, Australia; CI24RE(CA), CI422, or Hybrid L24) were used in 37 implantations, and implants from MedEl (Innsbruck, Austria; Sonata 31 or 24, or Concerto 20, 24, 28, or 31) were used in 37 implantations in the adult group. Of the four children, two were implanted with an implant from Cochlear (CI24RE(CA) and CI422) and two were measured using a test electrode from MedEl, which is an 18 mm long simplified cochlear implant with three contacts in the cochlea.
eABR measurements All recordings were made during surgery directly after implant insertion. Two separate systems were used, one for recording and one for stimulation. The stimulation system used was the Cochlear or MedEl programming system, which uses the CI as a stimulator. The recording system was an evoked potentials (EP)
Prognostic value of electrically evoked auditory brainstem responses
system, which was triggered to record from the stimulation system. Two different recording systems were used at the clinic, the Otometrics Chart 200 (GN Otometrics, Taastrup, Denmark) and the GSI Audera (GSI, Minneapolis, USA). Recording needle electrodes were positioned at the Vertex (Act), C7 (Ref ), and at the hairline on the neck (ground). Three different electrodes were stimulated for each patient, one each in the low-, mid-, and high-frequency regions. The stimulation pulse width was 25 μs, and the current level was 235 CL for Cochlear devices and 30 μs and 1000 q.u. for MedEl devices. These conditions produced the same stimulus charge from both systems and were above most patients’ C-levels in their sound processor maps. The recording system averaged 1000 sweeps and was filtered by a low pass filter at 3 kHz and a high pass filter at 10 Hz (GSI Audera) or by a low pass filter at 5 kHz and a high pass filter at 5 Hz (Otometrics Chartr 200). Latencies were measured on the result from an alternating stimulation or as an average of the latencies from the positive and negative stimulations in cases where no alternating polarity was used. In most cases, the eABR responses were verified by inverting the polarity of the stimulation, which resulted in an inversion of the artifact without an inversion of the eABR waveform. Figure 1 shows a typical result from the different modes of stimulation. The most identifiable peaks (II, III, and V) were marked according to standard clinical procedure.
Waveform assessment A modified classification table as described by Gibson et al. (2009) and Walton et al. (2008) was used to classify the quality of eABR waveforms (Table 1). We chose to simplify the table by including only latencies and leaving out amplitude measures, because we stimulated only at a single level, and both latencies and amplitudes change according to the level of stimulation. Abbas and Brown (1988) and Kim et al. (2008) claim that amplitudes undergo most changes when the stimulation level is modified, whereas latency changes are relatively small. A score of 3 represents a superior eABR quality and 0 an absent eABR response. A total waveform score was constructed by adding the scores from the three measured electrodes (0–9).
Hearing evaluation MS-words (Svensk Talaudiometri, C-A Tegnér AB, Stockholm, Sweden, 1998) were used to test speech perception in a sound-treated booth in free field at 65 dB SPL with the loud speaker at 0° azimuth. Only postoperative MS-word scores with CI were analyzed. In patients with no speech recognition (0% in MS-word test) bi-syllabic word testing and the threedigit test (Svensk Talaudiometri, C-A Tegnér AB,
Cochlear Implants International
2015
VOL.
16
NO.
5
255
Lundin et al.
Prognostic value of electrically evoked auditory brainstem responses
Table 1 Implant evoked electrical auditory brain stem response waveform score; table modified from Gibson et al. (2009) and Walton et al. (2008) Presence of waves Score
eII (Y/N)
eIII (Y/N)
eV (Y/N)
3 2 1 0
Y N N N
Y Y N N
Y Y Y N
compare latencies between wave V at the three different locations (base, middle, apex), and Bonferroni adjustments for three repeated measures were made. The Wilcoxon signed-rank test was used to compare wave form scores for the three different locations (base, middle, apex), and Bonferroni adjustments for three repeated measures were made. Linear regression analysis was used to explore relationships between waveform scores, waveform V latencies, wave V–III interval and MS-words. For waveform V latencies and wave V–III intervals, Pearson’s correlation coefficient (r) was calculated and for waveform scores, Spearman’s rank correlation coefficient (rs) was calculated.
Results eABR latencies and assessment of waveform quality
Figure 1 Typical eABR waveforms. (A) Recordings from alternating, positive and negative stimulation modes and (B) recordings from positive and negative modes.
Stockholm, Sweden, 1998) were also performed. Speech perception data were collected at 6 months or 1 year after surgery.
Statistical analyses Statistical analyses were made using SPSS 22.0 software (IBM 1989, 2013). Paired t-tests were used to
Latencies for wave II, III, and V are displayed in Table 2. Latencies increased towards the base of the cochlea. Significant latency differences were found between wave V in the low-frequency region and the high-frequency region (**P < 0.01, 95% confidence interval of the mean difference 0.09–0.46 ms), and between wave V in the mid-frequency region and the high-frequency region (**P < 0.01, 95% confidence interval of the mean difference 0.06–0.33 ms). Wave V values for the mid-frequency and the low-frequency regions were the most robust waveforms. In 66 of 74 (89%) recordings in the low-frequency region, we were able to identify wave V. In eight patients, wave V could not be observed. In five of these cases, the measurement failed owing to technical reasons, in two cases, the measurement was not performed on
Table 2 Means, standard deviations (SD) and 95% confidence intervals for eABR waves II, III and V. Latencies in ms. Ninety-five percent confidence interval for means in parentheses Low-frequency region Mean
SD
Wave II latency (ms) 1.42 (1.36–1.48) 0.17 Wave III latency (ms) 2.17 (2.11–2.22) 0.21 Wave V latency (ms) 3.99 (3.90–4.07) 0.36
256
Mid-frequency region
High-frequency region
n
Median
Mean
SD
n
Median
Mean
SD
n
Median
32
1.42
1.45 (1.35–1.54)
0.25
28
1.42
1.46 (1.31–1.60)
0.23
12
1.43
57
2.12
2.23 (2.17–2.29)
0.22
56
2.20
2.28 (2.17–2.39)
0.25
22
2.29
66
3.98
4.15 (4.06–4.24)
0.35
64
4.14
4.32 (4.17–4.48)
0.41
29
4.42
Cochlear Implants International
2015
VOL.
16
NO.
5
Lundin et al.
Prognostic value of electrically evoked auditory brainstem responses
Table 3 Mean scores and mean wave V–III differences; 95% confidence intervals for mean in parentheses.
Mean score (0–3) Median score (0–3) SD n Mean wave V–III (ms) Median wave V–III (ms) SD n
Low-frequency region
Mid-frequency region
High-frequency region
2.3 (2.14–2.51) 2 0.8 67 1.78 (1.70–1.86) 1.79 0.28 57
2.1 (1.92–2.34) 2 0.9 69 1.89 (1.82–1.96) 1.86 0.25 56
1.0 (0.66–1.25) 0 1.2 62 1.94 (1.82–2.06) 2.00 0.26 20
those electrodes, and in one case, wave V was not detectable. Therefore, in all but 1 of the 67 successful recordings (99%), wave V was detectable for the lowfrequency region. In the mid-frequency region, we identified wave V in 64 of 74 (86%) recordings. In 10 cases, wave V was not observed. In five of these cases, the measurement failed owing to technical reasons, and in the other five cases, wave V was not detectable. Therefore, in all but 5 of the 69 successful recordings (93%), wave V was detectable for the middle region. The waveform quality as assessed by a score from 0 to 3, was highest at the low- and midfrequency regions of the cochlea (mean scores 2.3 and 2.1, respectively) compared with 1.0 in the highfrequency region. There was a statistically significant difference between the waveform scores for the low and mid-frequency regions of the cochlea compared with the high-frequency region (***P < 0.001). Wave V–III latency increased towards the high-frequency region (Table 3). It was not possible to fully evaluate how the difference in electrode lengths contributed to
the waveform score, because the number of patients in each implant length group varied essentially. The mean waveform score was highest for the longest electrode (31 mm) though.
Hearing evaluation There were no correlations between MS-word scores and wave V latency in the three different regions (low r = −0.1 p = 0.4, mid r = −0.2 P = 0.07, high r = −0.1 P = 0.6) or wave V–III intervals in the three different regions (low r = −0.05 P = 1.0, mid r = −0.2 P = 0.2, high r = 0.04 P = 0.9) using linear regression analysis (Fig. 2). We also did not find any correlation between MS-word scores and the total score (rs = 0.2, P = 0.3) or the scores in the three different regions (low rs = −0.07, P = 1.0, mid rs = 0.06, P = 0.7, high rs = 0.2, P = 0.3) using linear regression analysis. No adjustments for repeated measures were performed, because no correlations were found.
Figure 2 Correlation between wave V latencies and monosyllabic (MS) words with CI. (A) Low frequencies (r = −0.1, P = 0.4) (B) mid frequencies (r = −0.2, P = 0.07) (C) high frequencies (r = −0.1, P = 0.6) and correlation between wave V–III interval and MSwords with CI (D) Low frequencies (r = −0.05, P = 1.0) (E) mid frequencies (r = −0.2, P = 0.2) (F) high frequencies (r = 0.1, P = 0.6).
Cochlear Implants International
2015
VOL.
16
NO.
5
257
Lundin et al.
Prognostic value of electrically evoked auditory brainstem responses
Table 4 Patients with absent wave V in the middle and/or apex regions and patients with no speech discrimination (MS-words, bi-syllabic words and three-digit-test)
Patient no.
Wave V (lowfrequency region)
Wave V (midfrequency region)
Total wave form score (0–9)
Patients with absence of wave V in the middle and/or apex region 12 No No 0
22
Yes
No
2
31
Yes
No
1
41
Yes
No
2
55
Not measured
No
–
Patients with no speech discrimination 9 Yes Yes
9
12
No
No
0
48
Yes
Yes
5
49
Yes
Yes
6
55
Not measured
No
–
74
Yes
Yes
4
Cochlear Implants International
2015
VOL.
16
NO.
Patient has very high levels in the map postoperatively. Detects only sound, no word discrimination. Acoustic neuroma in the CI-ear. No speech discrimination. Implant: Concerto28 (28 mm) Patient with otosclerosis. Sudden loss of hearing. Contralateral ear deaf for 30 years. Patient in coma for five weeks owing to Legionella one year before CI-surgery. Twentyfive percent on auditory only three-digit-test. Implant: CI422 (25 mm) Patient with otosclerosis and high levels in the map postoperative. Twenty-four percent on MS-word test. Implant: CI422 (25 mm) Young patient (18 years old at surgery) with high-frequency deafness. Fifty-eight percent on MS-word test. Implant: Concerto24 (24 mm) Patient reports mainly a feeling from implant stimulation postoperatively. Long-time deafness (50 years) before surgery. No speech discrimination. Implant: Concerto31 (31 mm) Deaf for 72 years prior to CI. CI use gives severe tinnitus. Non-user. Implant: Concerto31 (31 mm) Patient has very high levels in the map postoperatively. Detects only sound, no word discrimination. Acoustic neuroma in the CI-ear. Uses CI full time for environmental sounds. Implant: Concerto28 (28 mm) Reports sound and a feeling from implant use. Deaf for 37 years in the CI-ear prior to CI. Non-user. Implant: CI422 (25 mm) Severe prelingual hearing loss. 50 years old at implantation. Uses CI full time for environmental sounds. Implant: CI422 (25 mm) Patient reports mainly a feeling from implant stimulation postoperatively. Deaf for 50 years prior to CI. Uses CI but considers stopping using it. Implant: Concerto31 (31 mm) Not measured. Do not show up to appointments. User? Implant: Hybrid L (17 mm)
Table 4 shows the cases in which wave V was not detected for reasons others than technical failures at the mid- and/or low-frequency regions, and the cases in which patients had no open-set speech discrimination on any of the performed speech tests. In addition, the electrode lengths are shown in Table 4. A negative outcome was defined as a patient with no recordable speech discrimination with CI (measured as scoring 0% on MS-words or less than 10% on bi-syllabic words or less than 25% on the three-digit test). The results showed that if the patient has no detectable wave V in the low-frequency region, the outcome will most certainly be unsuccessful. Patients with no speech perception with CI had eABR wave V in the low-frequency region in 75% of the cases. One remarkable case with a total waveform score 9 and deafness duration before CI surgery of 72 years received hearing sensations from the implant (no speech perception) but could not use it, because the implant induced severe tinnitus.
258
Comment
5
Children with cochlear abnormalities Among the four children, no eABR waveforms could be seen from any of the stimulations (total score 0). Here, stimulation levels were increased to determine that no waveforms occurred at higher levels. One child tested via a Nucleus CI24RE(CA) implant received an ABI during the same surgery. In one child, the surgery proceeded with the intention to implant an ABI but had to be terminated for surgical reasons. One child did not receive an implant as parents and medical staff decided before surgery in case eABR measurements were negative. One child was implanted and monitored by a Nucleus CI422 implant and programmed 4 weeks after surgery. The child used the implant for 6 months but did not show any reactions to sound even though levels in the CI were set at the maximum.
Discussion Although interference and electrical noise from other surgical equipment may sometimes occur, eABR
Lundin et al.
recordings via the CI are relatively straightforward during surgery in most patients. Recording electrode placement and marking/coupling of the lead-wires to the EP equipment may be crucial factors. We found that wave V for mid-frequency and lowfrequency regions were often robust and wave V for the low-frequency region were recorded in every case where outcomes was successful, i.e., in patients benefiting from the implant. However, those waveforms were also obtained in a few cases when outcome was not successful (Table 4). Walton et al. (2008) described two children with auditory neuropathy spectrum disorder (ANSD) that upon MRI showed absence of auditory nerve but still presented recordable eABR responses via the CI. Both children had a poor CI outcome. In the three cases, in our study where no eABR-waveform responses were obtained ( patient no. 12 and 55, Table 4) and the child that received an implant despite absent intra-operative eABRs responses, the CI outcomes were unsuccessful and patients did not demonstrate open-set speech discrimination. However, the two adult patients continued to use their implants and benefitted from receiving background sound and/or awareness from the implant. Recently, one additional patient who showed no eABR responses was also implanted with an unsuccessful outcome. One crucial question is if there is possible to benefit from CI despite no eABR responses. Jeon et al. (2013) found that patients with ANSD did not always display eABR responses but had good CI outcomes. Greisiger et al. (2011) investigated eight children with ANSD and in comparison with eight non-ANSD children found no differences in eABR amplitudes or waveforms between the two groups. These authors found patent wave III and V in all investigated patients. Walton et al. (2008) compared eABR waveform quality for children with and without ANSD. The children with ANSD had significantly worse CI outcomes as well as decreased waveform quality compared with the nonANSD children. They also found a significant correlation between waveform quality and CI outcomes. Our results show longer latencies and indistinct waveforms at the high-frequency region compared with those of the low-frequency regions of the cochlea are consistent with earlier studies (Firszt et al., 2002a; Gordon et al., 2007; Miller et al., 1993). Zimmerman et al. (1995) and Gordon et al. (2007) showed that humans with hearing loss display smaller populations of spiral ganglion cells in the cochlear base (high-frequency region) compared with the apex (low-frequency region). Because the number of patients using a specific electrode length varied extensively in this study, no statistical analysis was performed regarding the influence of different electrode lengths on waveform scores. The length, size, and
Prognostic value of electrically evoked auditory brainstem responses
shape of the cochlea also vary greatly between individuals. Erixon et al. (2009) concluded in their study of 73 adult cochleae that the length of the first turn could vary from 20.3 to 24.3 mm. We could observe though that the mean total waveform score was higher for the longest electrodes (31 mm). Furthermore, Gordon et al. (2007) showed that latencies shortened after a year of implant use for all waves (II–V) at both high- and low-frequency regions in 50 examined children with early onset of severe to profound deafness. Apical versus basal differences persisted with one exception; wave V–III interval differences decreased with longer implant use. Firszt et al. (2002a) found wave V–III intervals to be 1.62 ms across all electrodes (low, mid, high). They stated that eABR wave V–III intervals are shorter than acoustic ABR V–III intervals because electrical stimulation via the CI actually enhances neural synchrony. Our results showed longer wave V–III intervals compared with these authors (Firszt et al. 2002a): 1.89 ms for mid-frequency regions and 1.87 ms for all electrodes (low, mid, high). These values may be considered normal for acoustic ABR V–III intervals. One explanation for the longer interval in our study could be differences in interpretation of wave V. Wave V often consists of wave IV and wave V combined, and there is rarely a clear separation between these two waves. This lack of clear visualization makes wave V difficult to mark. Similar to Firszt et al. (2002b), we found no correlation between speech perception outcomes (MSwords) and waveform latencies or waveform quality. However, absent waveform predicted no or poor CI outcomes. Walton et al. (2008) and Gibson et al. (2009) found a positive correlation between waveform quality and the Melbourne speech perception score (MSPS). The MSPS consists of a seven-point scale and a minimum of score 4 is needed to develop speech perception. They assessed eABR responses on 22 electrodes, and the maximum total score obtained was 66. Walton et al. (2008) found that patients with eABR scores >56 are more likely to get a MSPS ≥ 4. This study assessed eABR only on three electrodes and the MS-word test was used to assess speech perception outcomes. The reason for measuring eABR on only three electrodes in this study was that measurements were an integral part of the regular clinical intra-operative CI test protocol, with the intention not to prolong surgery. One explanation for the different results could be the different speech perception measures used. Gallégo et al. (1998) found the wave V–III interval to be a strong predictor by testing 17 patients using a phoneme-test. They measured eABR at eight different levels to correct for the different postoperative levels required by patients. They also used an algorithm for automatic wave-latency measurement.
Cochlear Implants International
2015
VOL.
16
NO.
5
259
Lundin et al.
Prognostic value of electrically evoked auditory brainstem responses
Our study used a different speech perception score and would probably have been more accurate if we had been able to record from all electrodes at a number of different levels that could be compared at a later time as waveform latency decreases and amplitude increases as the stimulation levels are augmented. Battelino et al. (2009) performed eABR prior to CI using a golf club stimulating electrode close to the round window. They were able to record eABR in only 64% of their cases using that technique and measured CI outcome with pure tones audiometry after CI. In their 46 patients, a positive eABR strongly correlated with a positive CI outcome. We obtained similar results, because all patients who showed a positive eABR were able to detect sound through the implant although speech perception varied among these patients. Kubo et al. (2001) showed that the eABR amplitude growth curve correlated well with consonant recognition scores in cochlear implant patients 1 month following surgery, but when tested at 3, 6, and 12 months after implantation, the correlation was no longer apparent. They also concluded that the event-related potential (P300) latency was a better indicator for hearing ability after CI, meaning that the higher auditory system is of greater importance for speech perception with CI. To use a CI for eABR in the decision-making process between CI and ABI, as done for two of the four children in this study, may not be considered cost-beneficial. The reason for this was the uncertainty of the radiologist to establish the presence of a patent or hypoplastic auditory nerve. Therefore, it was considered necessary to first try a CI. An alternative was to perform round window electrode stimulation. Owing to particular conditions it was found better to proceed with a CI testing before continuing the ABI surgery. In the other two pediatric cases, eABRs were assessed using a test electrode from MedEl (Innsbruck, Austria). It consists of an 18 mm long basic CI array with three contacts inside the cochlea. Such a device may be more useful and cost-effective in future decision-making between CI and ABI. However, we believe that an even longer device than the current version might be beneficial. Not all measurements were performed in the alternating polarity in this study. This is a limitation as there is a slight shift in latencies between the different stimulation modes, even though the mean values were used in those cases. In addition, waveforms II, III and V were marked manually. A more accurate latency measure would have been obtained if we would have been able to stimulate at a number of different levels to adjust for the patients’ different levels used later on in their sound processor. Sometimes, eABR measurements were difficult to collect at the operating room for reasons such as interference during wound closure, etc.
260
Cochlear Implants International
2015
VOL.
16
NO.
5
Taken together, eABR monitoring seems to be a useful tool to further predict the outcome of CI surgery. In difficult cases with inner ear malformations as well as long-time deafness, eABR may be helpful to the surgeon and team to assess whether a patent or hypoplastic auditory nerve may provide sufficient electric information to the auditory cortex. However, several factors such as anatomical conditions, cognition, age at implantation, deafness duration, cause of deafness, and age at onset of deafness contribute to the final outcome and eABR responses cannot be used as sole predictors. However, we believe that this method gives important additional information and seems to have a definitive role in the diagnostic armamentarium to better consider complementary surgical strategies such as ABI.
Conclusion eABR via the CI is relatively straightforward during surgery in most patients. Wave V for the low-frequency regions was registered in all patients with successful CI outcomes. Patients with no eABR waveforms showed poor CI outcomes in terms of speech perception. eABR may be particularly useful in many cases to assist surgeons in strategically choosing between CI and ABI.
Acknowledgements The authors thank statistician Lars Lindhagen for advice on the data analyses.
Disclaimer statements Contributors KL is the corresponding author and has performed the measurements, analyzed the data, and written the text. FS has also performed the measurements and contributed to the writing and the analyzes. HR-A has done a massive contribution to the text. Funding This study was supported by grants from Gunnar Arnbrinks foundation and Uppsala läns landstings FoU-medel. Conflicts of interest The authors report no conflicts of interest. Ethics approval The study was approved by Uppsala Ethical Review Board (19/11-2014, Dnr: 2014/437).
References Abbas, P.J., Brown, C.J. 1988. Electrically evoked brainstem potentials in cochlear implant patients with multi-electrode stimulation. Hearing Research, 36: 153–162. Battelino, S., Gros, A., Butinar, D.A. 2009. Comparison of electroaudiometry (EAM) and electrically evoked brainstem response (eABR) with the cochlear implantation hearing results. International Advanced Otology, 5(1): 100–103. Cinar, B.C., Atas, A., Sennaroglu, G., Sennaroglu, L. 2011. Evaluation of objective test techniques in cochlear implant users with inner ear malformations. Otology Neurotology, 32: 1065–1074.
Lundin et al.
Erixon, E., Högstorp, H., Wadin, K., Rask-Andersen, H. 2009. Variational anatomy of the human cochlea: implications for cochlear implantation. Otology Neurotology, 30: 14–22. Firszt, J.B., Chambers, R.D., Kraus, N., Reeder, R.M. 2002a. Neurophysiology of cochlear implant users I: effects of stimulus current level and electrode site on the electrical ABR, MLR and N1-P2 response. Ear Hearing, 23(6): 502–515. Firszt, J.B., Chambers, R.D., Kraus, N. 2002b. Neurophysiology of cochlear implant users II: comparison among speech perception, dynamic range, and physiological measures. Ear Hearing, 23(6): 516–531. Gallégo, S., Frachet, B., Micheyl, C., Truy, E., Collet, L. 1998. Cochlear implant performance and electrically-evoked auditory brain-stem response characteristics. Electroencephalography Clinical Neurophysiology, 108: 521–525. Gibson, W.P.R., Sanli, H., Psarros, C. 2009. The use of intra-operative electrical auditory brainstem responses to predict the speech perception outcome after cochlear implantation. Cochlear Implants International, 10(S1): 53–57. Gordon, K.A., Papsin, B.C., Harrison, R.V. 2007. Auditory brainstem activity and development evoked by apical versus basal cochlear implant electrode stimulation in children. Clinical Neurophysiology, 118: 1671–1684. Greisiger, R., Tvete, O., Shallop, J., Elle, O.J., Hol, P.K., Jablonski, G.E. 2011. Cochlear implant-evoked electrical auditory brainstem responses during surgery in patients with auditory neuropathy spectrum disorders. Cochlear Implants International, 12(S1): 58–60.
Prognostic value of electrically evoked auditory brainstem responses
Jeon, J.H., Bae, M.R., Song, M.H., Noh, S.H., Choi, K.H., Choi, J.Y. 2013. Relationship between electrically evoked auditory brainstem response and auditory performance after cochlear implant in patients with auditory neuropathy spectrum disorder. Otology Neurotology, 34: 1261–1266. Kim, A.H., Kileny, R.R., Arts, P.R., El-Kashlan, H.K., Telian, S.A., Zwolan, T.A. 2008. Role of electrically evoked auditory brainstem response in cochlear implantation of children with inner ear malformations. Otology Neurotology, 29: 626–634. Kubo, T., Yamamoto, T., Iwaki, M., Matsukawa, M., Doi, K., Tamura, M. 2001. Significance of auditory evoked responses (eABR and P300) in cochlear implant subjects. Acta Otolaryngology, 121: 257–261. Miller, C.A., Abbas, P.J., Brown, C.J. 1993. Electrically evoked auditory brainstem response to stimulation of different sites in the cochlea. Hearing Research, 66: 130–142. Tysome, J.R., Axon, P.R., Donnelly, N.P., Evans, G. D., Ferner R.E., Fitzgerald O’Connor A.F. et al. 2013. English consensus protocol evaluating candidacy for auditory brainstem and cochlear implantation in neurofibromatosis type 2. Otology Neurotology, 34: 1743–1747. Walton, J., Gibson, W.P.R., Sanli, H., Prelog, K. 2008. Predicting cochlear implant outcomes in children with auditory neuropathy. Otology Neurotology, 29: 302–309. Zimmerman, C.E., Burgess, B.J., Nadol, J.B. 1995. Patterns of degeneration in the human cochlear nerve. Hearing Research, 90: 192–201.
Cochlear Implants International
2015
VOL.
16
NO.
5
261
